Anthropic now sits at the center of a roughly $200 billion financing network for chips and data centers. The company needs enormous computing power to develop and operate its AI systems. Google supplies much of that capacity, while Broadcom, banks, and investment firms help finance the supporting infrastructure.
As a result, Google does not fund the entire expansion itself. Instead, several participants finance different parts of the buildout. This structure gives Anthropic access to computing resources while distributing the cost across multiple partners.
Anthropic Is Building a Massive Compute Pipeline
Anthropic has committed to spend $200 billion with Google Cloud over five years. The agreement covers cloud services and computing capacity. On another end, the company will access about 3.5 gigawatts of Google TPU capacity through Broadcom from 2027.
Additionally, Apollo and Blackstone also support a $35 billion Broadcom financing platform for AI computing. Together, these arrangements secure future capacity for AI expansion while dividing equipment and funding requirements among several participants.
AI Demand Is Outgrowing Traditional Infrastructure Funding
However, chips represent only one part of the challenge. AI development also requires electricity, networking equipment, land, and data centers. Morgan Stanley estimates global data-center construction could require about $2.9 trillion through 2028.
Yet, hyperscalers cannot cover the entire bill through internal cash flows. Because of this, banks, private credit firms, and other investors have an expanding role in funding new facilities.
As a result, companies can combine leases, project financing, guarantees, and outside investment. Suppliers can secure major orders, while investors finance physical assets needed for deployment.
Meanwhile, AI companies gain access to facilities without paying every construction cost upfront. Long-term commitments can also give investors clearer visibility into future demand.
The Risk Is Moving Beyond Corporate Balance Sheets
More importantly, spreading financing does not eliminate financial risk, it changes who is at risk. For example, a $15 billion Texas data-center project was proposed. Morgan Stanley leads financing for a campus serving Anthropic while Google would guarantee billions in lease and power obligations. In addition, Google could also receive an equity stake in the project.
Therefore, Google can support the expansion without directly funding the entire facility. However, guarantees still create financial exposure. Lenders and investors also assume risks tied to the project. Anthropic must maintain sufficient business to meet its commitments.
AI Infrastructure Financing Is Entering Its Next Phase
Looking ahead, Anthropic could become part of a broader financing model for AI infrastructure. Morgan Stanley expects credit markets and private capital to fund a growing share of the industry’s expansion. Additionally, Broadcom’s platform shows how chip suppliers can work with institutional investors.
As demand rises, more projects could combine chip supply, customer commitments, outside financing, and dedicated facilities. The approach could reshape how companies build expensive computing systems. For Anthropic, reliable access to computing remains essential. For investors, the opportunity lies in financing the physical foundation supporting AI growth.
